A new technique simulates viscoelastic fluids like honey interacting with solid and deformable objects. The method uses two parameters to control relaxation time and compliance, and handles scenes with over 100,000 particles at six frames per second.
Dear fellow scholars, this is two minute papers Withehir. It has been a while now since we've talked about fluid simulations, and now it is time for us to have a look at an amazing technique that creates simulations with viscoelastic fluids plus rigid and deformable bodies. This is a possibility with previous techniques, but takes forever to compute and typically involves computational errors that add up over time and lead to a perceptible loss of viscoelasticity. I'll try to explain these terms and what's going on in the moment. There will be lots of eye candy to feast your eyes on throughout the video, but I can assure you that this scene takes the cake. The simulation phase here took place at one frame per second, which is still not yet fast enough for the much coveted real time applications, however, is super competitive compared to previous offline techniques that could do this. So what does viscosity mean? Viscosity is the resistance of a fluid against deformation. Water has a low viscosity and rapidly takes the form of the cup we pour it into, where honey ketchup and peanut butter have a higher viscosity and are much more resistant to these external forces. Elasticity is a little less elusive concept that describes to what degree a piece of fluid behaves like elastic solids. However, viscous and elastic are not binary yes or no concepts. There is a continuum between the two, and if we have the proper machinery, we can create viscoelastic fluid simulations. The tau parameter that you see here controls how viscous or elastic the fluid should be and is referred to in the paper as relaxation time. As tau is increased towards infinity, the friction dominates the internal elastic forces and the polymer won't be able to recover its structure. Well, the opposite cases where we reduce the tau parameter to zero, then the internal elastic forces will dominate and our polymer will be more rigid. The alpha parameter stands for compliance, which describes the fluidity of the model. Using a lower alpha leads to more solid behavior and higher alphas lead to more fluid behavior. The cool thing is that as a combination of these two parameters, we can produce a lot of really cool materials ranging from viscous to elastoplastic to inviscid fluid simulations. Have a look at this honey pouring scene. Mmm. This simulation uses more than a hundred thousand particles and six of these frames can be simulated in just one second. Wow. If we reduce the number of particles to a few tens of thousands, real time, human interaction with these simulations also becomes a possibility. A limitation of this technique is that most of our decisions involving the physical properties of the fluid are collapsed into the tau and alpha parameters. If we are looking for more esoteric fluid models, we should look elsewhere. I am hoping that since part of the algorithm runs on the graphics card, the speed of this technique can be further improved in the near future. That would be awesome. Admittedly, we've only been scratching the surface, so make sure to have a look at the paper for more details. Thanks for watching and for your generous support. And I'll see you next time, Ra.
